HP (Hewlett-Packard) 9000 V2500 SCA ユーザーズマニュアル

ページ / 206
Chapter 7
147
Recovering from failures
Abnormal system shutdowns
Abnormal system shutdowns
Abnormal systems shutdowns (often referred to as system crashes) can 
occur for many reasons. In some cases, the cause of the crash can be 
easily determined. In some extreme cases, however, it may be necessary 
to analyze a snapshot (called a core dump or simply dump) of the 
computer’s memory in order to determine the cause of the crash. This 
may require the services of the Hewlett-Packard Response Center.
V-Class servers using HP-UX Release 11.0 or greater employ a more 
efficient dump mechanism than other HP servers using previous releases 
of HP-UX. This mechanism is called fast dump.
Fast dump
When a system crashes, the operator can now choose whether or not to 
dump, and if so, whether the dump should contain the relevant subset of 
memory or all memory (without operator interaction).
By default fast dump selectively dumps only the parts of memory that 
are expected to be useful in debugging. It improves system availability in 
terms of both the time and space needed to dump and analyze a large 
memory system.
The following commands allow the operator to configure, save, and 
manipulate the fast core dump:
crashconf
—Configures the destination and contents of a crash 
dump without rebooting. See the crashconf(1M) man page for more 
information.
savecrash
—Runs at boot time and saves any information that may 
be overwritten by normal system activity. See the savecrash(1M) man 
page for more information.
crashutil
—Saves or manipulates the crash dump (if desired). It can 
format the dump snapshot so that it can be read by the older 
commands. See the crashutil(1M) man page for more information.
Installations that used to call 
savecore
 in any way other than by the 
HP-supplied, unmodified 
/sbin/init.d/savecore
 script need to be 
updated to use 
savecrash
 and/or 
crashutil
.